The error message was always there, but recently we started to log keepalived-state-change messages to syslog, at which point it started to trigger monitoring tools. There is no known functional deficiency when the error happens. It occurs when the router is deleted, and the -change daemon receives a SIGTERM. At this point, it's not a problem that the daemon doesn't shuts down gracefully, since the process has nothing to clean up after itself anyway.
I agree that this should be solved, but lowering the priority/severity to reflect the description above.